Earlier quoted context omitted.
Fascinating.. you are aware of the context of how WYSIWYG came about? So presumably you don't want to return to that state, where you literally would have no idea what it would look like until several minutes later when it finally came out on the printer? Can you explain a little bit further what your ideal paradigm is?
Ideally people would write mark-up in text editors. If you want to be very nice, I guess it would be OK to have drag-and-drop WYSWYG environments that spit out the markup code, but there should be a very high priority on making sure the code produced is human-readable.
Case in point: my book was edited in Vellum, and it can generate PDF and EPUB. The PDF had a widow (one line at the top of the last page of the chapter), and Vellum just omitted that one-line page. To the reader it seemed like a typo (which it was, in a sense).
I "fixed" it by removing a few words up above it, so everything fit on the last full page. But it was only that I knew about widow/orphan control that I could figure that out. Just imagine how much trouble it would be to make things perfectly WYSIWYG on every printer and every type of document.